DOI: 10.5152/tepes.2024.

23024

RESEARCH ARTICLE

Modeling and Simulation of a Series and Parallel Battery Pack Model

in MATLAB/Simulink
Azemtsop Manfo Theodore , Mustafa Ergin Şahin
Department of Electrical and Electronic Engineering, Recep Tayyip Erdoğan University, Rize, Turkey

Cite this article as: A. M. Theodore and M. E. Şahin, “Modeling and simulation of a series and parallel battery pack model in MATLAB/simulink,” Turk J
Electr Power Energy Syst., 2024. [epub ahead of print]

ABSTRACT

Lithium-ion batteries have recently become the focus of research in vehicle applications due to their numerous advantages. Lithium-ion batteries have higher
specific energy, better energy density, and a lower self-discharge rate than other secondary batteries, making them appropriate for electric vehicles and hybrid
electric vehicles. Nonetheless, worries about safety, cost, charging time, and recycling have hampered the commercial usage of lithium-ion batteries for auto-
motive applications. An accurate battery model on a simulation platform is required for the development of an effective battery system. In this study, a battery
model is built in MATLAB/Simulink. Two variations are available: one with a series–parallel battery arrangement and a single model without configuration.
The structure of the proposed model is provided and explained in detail. Based on the test results, the developed battery model was validated. A comparison
shows that the model created can accurately predict current, voltage, and power performance. This model is designed for Eaton Electromechanical Battery
Li-ion 18650 batteries but is also said to work with other types of batteries. The simulation takes into account the battery’s state of charge, current, voltage,
and power requirements.

Index Terms— Charge status, hybrid electric vehicles, lithium-ion battery, MATLAB/Simulink

I. INTRODUCTION have constraints such as slow charging and a limited lifespan

The energy factor is a vital component of the growth of any soci- [4]. LIBs, in particular, have become an essential component
ety [1]. Battery technology has advanced significantly in recent of decentralized, often tiny or micro-scale, off-grid renewable
decades, particularly in lithium-ion batteries (LIBs) [2]. LIBs have energy systems, supplanting diesel generators in many remote
dominated the market for advanced energy sources in today’s areas. LIBs are a potentially clean technology that can be utilized
society since they are widely employed in several industries to replace typical fossil-fuel-powered gadgets. LIBs with better
such as electronics, electric vehicles (EVs), and energy stor- specific energy, low self-discharge, and improved coulombic effi-
age systems [3]. A rechargeable battery may perform an elec- ciency are required for electric cars, railways, and spatial tech-
trochemical conversion, which means that the stored chemical nologies [5].
energy can be turned directly into electric energy. In the sys-
tem, chemical energy is transformed into electric energy during Accurate battery information, such as state of charge (SOC), current,
the charge and back to chemical energy during the discharge. and voltage, is crucial for circuit designers to regulate the energy
Depending on the electrochemical system, this system may be consumption of battery electrical systems. As a result, having a reli-
charged and discharged more than 500 times. Because lithium- able battery model is critical for forecasting battery properties during
ion movement within the battery is followed by charge flow in the circuit design process. The SOC of LIBs indicates the remaining
an external circuit, the efficiency of lithium-ion movement in power of the battery [6]. It can be used to prevent the battery from
the electrolyte impacts the battery capacity [2]. Rechargeable being overcharged or overdischarged, predict the driving range of
batteries are important for future technologies because they EVs, and eliminate the inconsistency of different cells. Furthermore,
have the potential to be employed as energy storage elements in the estimation method is sensitive to factors like temperature, cycle
green technology applications such as EVs and photovoltaic sys- time, discharge rate, voltage, and noise, making it difficult to cor-
tems. Two traditional energy storage systems, such as batteries, rectly estimate the SOC of a battery in real time [7].

Received: September 28, 2023

Corresponding author: Azemtsop Manfo Theodore, azems​ouley​mane@​yahoo​.fr
Accepted: November 10, 2023
Publication Date: January 3, 2024
Content of this journal is licensed under a Creative Commons
Attribution-NonCommercial 4.0 International License.
 TEPES 2024 [epub ahead of print]
Theodore and Şahin. Modeling and Simulation of a Series and Parallel Battery Pack Model

A comparable circuit model is commonly used by circuit designers nominal battery voltage in an LIB is 3.7 volts per cell. When an elec-
since it can be easily performed in circuit simulators [8, 9]. To predict trical device is linked to an LIB, the blocked electrons travel through
the current–voltage behavior of the batteries, the simulation mod- the device and power it. An LIB cell can function as a power cell
els employ the MATLAB software curve-fitting toolbox, which con- (which delivers a high current load for a short period) or an energy
nects the circuit model to the resistor–capacitor circuit [10]. A circuit cell (which delivers sustained current for a long period of time). An
version similar to the battery may be created within the MATLAB/ LIB is made up of an anode, a cathode, an electrolyte, a separator,
Simulink environment to series the battery equations using differen- and a current collector [18]. Table I outlines the properties of these
tial equations that calculate the dynamics of the battery parameters various battery technologies [14].
[11-13].
B. Classification of Lithium-Ion Batteries
The proposed LIB circuit model is built in MATLAB/Simulink in this The vocabulary used to describe LIB chemistry is denoted by
study. Unlike the single-battery model, the recommended battery shortened letters. The chemistry of cathode materials impacts
architecture’s parameters vary based on SOC, current, voltage, and the efficacy of an LIB. Cobalt has long been used as a compo-
capacity, allowing the circuit designer to set the parameters primar- nent of LIBs. The difficult acquisition of expensive cobalt, on
ily based on battery conductivity. Furthermore, the proposed ver- the other hand, calls into question its usefulness as a battery
sion is without difficulty related to different MATLAB/Simulink circuit material. Battery manufacturers have been researching several
blocks, bearing in mind real-time SOC estimation. As a result, accu- types of LIBs to prevent risky sourcing methods while boosting
rate SOC estimation is necessary to protect the battery and avoid cost, loading capabilities, and longevity. However, there are sev-
overcharging and discharging. This will also help to prolong the life eral types of LIBs, each with its own set of advantages, such as
of the battery. lithium manganese oxide, which has a high specific power; lith-
ium cobalt oxide, which has high specific energy; lithium nickel
II. LITERATURE REVIEW cobalt aluminum (NCA) oxide and lithium nickel manganese
A. Electrochemical Storage Device oxide, which are the most thermally stable and cheapest; and
An electrochemical generator generates electrical energy directly lithium titanate oxide, which has a fast charge, a long life, but a
from chemical energy. Batteries, accumulators, and fuel cells are the higher cost and low specific energy [19]. The chemical composi-
three types of electrochemical generators. The history of electro- tion of an LIB is crucial to its power, resilience, and safety in a
chemical generators begins in 1800, with Alessandro Volta’s inven- variety of applications. Tesla, for example, uses a lithium-iron-
tion of the non-rechargeable primary cell [14]. phosphate battery chemistry for its standard-range vehicles and
an NCA battery chemistry for its longer-range vehicles. Each bat-
Gaston Planté constructed the lead accumulator in 1859 after dis- tery performs differently and is chosen depending on the appli-
covering the reversibility of electrical chemical reactions due to cation [20]. Table II depicts commercial LIBs and their various
current reversal. It is composed of lead alloy grids that have been features [21].
glued together with a mixture of sulfuric acid, lead oxide, and
water as the active ingredient. This strategy enabled an electric car C. Battery Structure and Modeling
shaped like a torpedo to surpass 100 km/h in 1899 [15]. It was the An LIB pack is composed of clusters of individual LIB cells that are
first commercially available rechargeable battery. Lead-acid batter- organized in series and parallel, or both directions to generate the
ies are still utilized in automobiles for 12V and 15V power sources. desired capacity, power density, or voltage for a variety of applica-
The popularity of this type of battery can be attributed to the inex- tions. A battery has an extremely short cycle life when exposed to
pensive cost of lead and sulfuric acid, ease of manufacture, and dampness. As a result, it is vital to monitor moisture, cold tempera-
short lifespan [16]. tures, and increasing storage temperatures, as these factors can
have a major impact on battery performance. A battery manage-
Sony introduced the first lithium-ion rechargeable batteries in 1991 ment system (BMS) is an electrical system that monitors individual
[17]. This technology quickly rises to prominence due to its specific cells within a battery pack to improve safety and performance.
energy, load capacity, and electromotive force performance. The
LIBs do not refer to any specific device type. Instead, there are a
variety of chemical and mechanical arrangements, each with its
Main Points
own set of features [22]. A BMS is required to properly employ a
• A new series/parallel lithium battery pack model was pro- battery system in a real-world application. BMSs are used to moni-
posed using MATLAB/Simulink. tor and control numerous cell states as well as to activate in the
• The characteristics of the proposed battery model were sim- event of abuse or failure [23, 24]. In recent years, the development
ulated and analyzed. of control-oriented models has resulted in a plethora of modeling
approaches that can be broadly characterized as electrochemical
• The discharging behavior in a single battery and the pro-
posed pack model. based models, equivalent circuit models, and data-driven models
[25, 26].
• The performance of the proposed model for application in
electric vehicles and hybrid electric vehicles was compared
With a better understanding of the battery’s behavior, the next
and evaluated.
step in its management is the estimation of unmeasured internal
 TEPES 2024 [epub ahead of print]
Theodore and Şahin. Modeling and Simulation of a Series and Parallel Battery Pack Model

TABLE I.
COMPARISON OF SEVERAL BATTERY TECHNOLOGIES

Electromotive Number of Charging Gravimetric Energy Volumetric Energy

Type Force in Volts Cycles Efficiency (%) Density (Wh/kg) Density (Wh/L) Temperature (°C)

Lead acid 2.1 500–1200 – 15–45 40−80 (−40)–40

Nickel-cadmium 1.2 ≈2000 60 30–60 80 (−20)–60

Nickel metal hydride 1.2 500–1200 60 100 200 (−20)–60

Lithium polymer 3.7 ≈1000 – 100–130 140–435 (−40)–40

Lithium ion 3.6 1000–10 000 95 150 300 (−20)–60

TABLE II.
LITHIUM-ION COMMERCIAL BATTERY COMPARISON

Abbreviation LCO LNO LMO NMC LFP NCA LTO

Year Since 1991 Since 1996 Since 1996 Sine 2008 Since 1993 Since 1999 Since 2008

Nominal voltage (V) 3.7~3.9 3.6~3.7 3.7~4.0 3.8~4.0 3.2~3.3 3.6~3.65 2.3~2.5

Specific energy (Wh/kg) 150~200 150~200 100~150 150~220 90~130 200~260 70~85

Charge (C) 0.7~1 0.7~1 0.7~1 0.7~1 1 0.7 1

Discharge (C) 1 1 1 1 1 1 10

Lifespan 500~1000 >300 300~700 1000~2000 1000~2000 500 3000~7000

Thermal runaway (°C) 150 150 250 210 270 150 –

LCO, lithium cobalt oxide; LFP, lithium-iron phosphate; LMO, lithium manganese oxide; LNO, lithium nickel dioxide; LTO, lithium titanate oxide; NCA, lithium nickel cobalt
aluminum oxide; NMC, lithium nickel manganese.

variables. State of charge and state of health are critical cell param-
eters that cannot be directly quantified [27]. Battery parameters play
an important role in battery management. While they change slowly
(due to aging), learning more about them is desirable. While offline
testing or open-loop calculation (for example, Coulomb counting for
SOC calculation [28]) can provide an estimate, live closed-loop esti-
mation of these indicators is crucial for robust and accurate real-time
monitoring.

In the MATLAB graphical editor Simulink, a generic model of an LIB

is built and validated (Fig. 1). It is portrayed as a controlled volt-
age source that is dependent on the SOC of the battery. The con-
cept behind this design is to use a simple way to extract the input
parameters for the battery model (shown in Fig. 2) from the battery
manufacturer’s catalog data. The model has different battery voltage
dependency features depending on the operation modes.

III. LITHIUM-ION BATTERY MODEL

These days, the most common battery is LIB, which may be discov-
ered in most transportable gadgets, starting from mobile telephones
to laptops. Fig. 1 depicts a photo of this battery. LIBs have an exten-
sive operational temperature range, as they can rate between −20°C
and 60 °C and discharge between −40°C and 65 °C. LIB cells have a
voltage of 2.5 to 4.2 V, which is about three times better than differ- Fig. 1. Li-ion battery model [31].
ent styles of batteries [29–31].
 TEPES 2024 [epub ahead of print]
Theodore and Şahin. Modeling and Simulation of a Series and Parallel Battery Pack Model

A. Mathematical Modeling of the Battery

This is accomplished by applying a mathematical battery model
to calculate the open circuit voltage (OCV) of the battery and SOC
parameters. Peukert’s equation is one of the simplest mathemati-
cal equations for estimating the SOC of a battery. Peukert’s rule was
published in 1897 by the German physicist W. Peukert to define bat-
tery capacity variations throughout discharge time using Peukert’s
constant [32]. The capacity of a lead-acid battery is calculated using
the rate of discharge [33]. The discharge curve of the battery in Fig. 3
is non-linear and is expressed by (1), where the charge current of the
Shepherds model [34] is positive (i.e., >0):

Q * Q
f1 (it , i * , i) = E0 - K .i - K .it + Aexp ( -B.it ) (1)
Q - it Q - it

where E0 is the constant voltage (V), K is the bias resistance (Ω), i* is

the low-frequency current dynamics (A), it is the battery current (A),
Q is the maximum battery capacity (Ah), A is the exponential voltage,
and B is the exponential capacity (Ah−1).

The reverse discharge curve is depicted here (beginning with an

empty battery, progressing to a quick rise to rated voltage after
charging to rated voltage, and finally to the exponential area when
the OCV is restored) (2):

Q Q
f1 (it , i * , i) = E0 - K .i * - K .it + Aexp ( -B.it ) (2)
Q + 0.1it Q - it

Because the load current has the opposite sign (i* ˂ 0), the bias resis-
tance changes, and the function of the load voltage is slightly altered
[35]. The effect of temperature on the LIB operation is not consid-
ered in this paper.

As the battery discharge curve in Fig. 3 shows, battery performance

decreases as the discharge rate increases. The discharge rate ampere
value is expressed as (3):

C p = I kt (3)

Fig. 2. Proposed model of series/parallel lithium-ion batteries packed

model. Fig. 3. Typical battery discharge curve.
 TEPES 2024 [epub ahead of print]
Theodore and Şahin. Modeling and Simulation of a Series and Parallel Battery Pack Model

where Cp is the capacity at a discharge rate of 1 amp (Ah), I is the

current drawn by the load, and t is the time (h). Normally, the bat-
tery cell discharge rate is not 1 Ah. Changing the calculation formula
according to the capacity and discharge amount gives (4):

k
æC ö
Cp = H ç ÷ (4)
è IH ø

In (4), the nominal discharge time in hours is H, the nominal power

Ah at discharge rate is C, I is the measured discharge current, k is the
Peukert constant, and t is the battery discharge time in hours [36].

IV. BATTERY DISCHARGE EXPERIMENTS

A. Voltage Versus Time
In EVs, the number of batteries is connected in a series-parallel
configuration to meet the load requirement. Due to manufacturing
methods, not all batteries reach maximum voltage at the same time
during charging. This condition causes voltage imbalances between
batteries and, as a result, lesser capacity from the entire battery
string [37]. Generic batteries and their standards have undergone
modifications to accommodate the specific requirements of differ-
ent battery types and their charge and discharge characteristics. By
utilizing the default properties of the battery model, we employed
the draw option to visually depict the parameter values. An Eaton
Electromechanical Battery (EEMB) Li-ion 18650 battery serves as the
designated power source for reference purposes. The positive (+)
and negative (−) terminals of the battery are utilized as buffer termi-
nals, while multiple vector signals (m) [38] are employed as output
terminals.
Fig. 4. (a) Block parameters and (b) battery discharge characteristics.

Fig. 4 presents the (Fig. 4a) block parameters and (Fig. 4b) typical
discharge characteristics of an LIB. The features can be classified
B. Battery Voltage Versus Discharge Capacity
into three primary categories. The first category is the exponential
The LIB model is stimulated with electrical aspects to estimate bat-
range, which encompasses voltages exceeding the nominal value.
tery frequency at room temperature (RT). A continuous discharge
The second category is the battery operating point, which refers to
test is performed on the battery with a constant current. Typical dis-
the period when the battery is inactive and generates a consistent
charge curves for the EEMB Li-ion 18650 battery model operated at
discharge current.
different discharge currents (6.5 A, 13 A, 26 A, and 52 A) are shown
in Fig. 5. During battery testing, the battery voltage is set to a limit
In the nominal operating range of the battery, there is a slight varia-
value (e.g., 2.775 V) defined by the battery manufacturer to avoid
tion in voltage during the discharge process. When the battery is
lasting damage to the battery.
scheduled for discharge, there is a third operational period in which
the battery undergoes the process of discharging. The battery dis-
charge curve, depicted by the blue line in Fig. 4, exhibits nonlinearity A noticeable trend is observed wherein the discharge capacity dimin-
and is characterized by the Shepherd model [34]. The current flowing ishes with an increase in the applied current. The discharge curve
during the charging process is positive (i* > 0). The polarization volt- exhibits a downward shift when the operating current is increased
age, which can be determined by multiplying the series resistance by from 6.5 A to 52 A. According to the data presented in Fig. 5, the bat-
the battery current, serves as an indicator of the battery’s discharge tery’s full charge voltage at RT is recorded as 4.30 V. However, when
behavior. Fig. 4 illustrates the alterations in discharge characteristics the battery is discharged at various currents, the discharge capacity
resulting from variations in current. decreases to 52 A, resulting in a voltage drop of approximately 5%
to 4.20 V. The development of a battery performance management
In the voltage–time diagram, the current value of 5.4348 A repre- system is highly significant in this context.
sents the discharge characteristics observed during discharge at a
constant current value. Additionally, the battery discharge was found V. A GENERIC MODEL DISCHARGE AND CHARGE
to be correlated with currents of 0.52 C (6.5 A), 1.04 C (13 A), 2.08 C The basic configuration of an LIB includes a current measurement
(26 A), and 4.16 C (52 A). The discharge duration is 2 h at a rate of 0.5 connected to the battery’s positive terminal, a load resistance (R)
C and 15 min at 4 C. As a consequence, the battery’s range experi- in parallel with a voltage measurement connected to the battery’s
ences a notable decrease as the rate of current discharge rises. negative terminal, and the battery circuit’s internal resistance.
 TEPES 2024 [epub ahead of print]
Theodore and Şahin. Modeling and Simulation of a Series and Parallel Battery Pack Model

Fig. 5. Estimation of the battery voltage versus discharge capacity.

Resistance limits or regulates the current flow within the battery æ V ö V2

model to prevent overcharging. The output signal parameter is not P = V .I = RI 2 = V . ç ÷ = (5)
èRø R
used if the block is not used in pointer simulation. The simulating
cursor is activated by a Powergui block set up in the model. The auto
where V is the supply voltage applied, I is the battery current, and R
solver is not selected in the Simulink solver options to enable the
is the resistance.
Powergui solver. The simulation type, including simulation param-
eters and settings, is defined by the options menu block Portable
The power output P of the battery pack model is calculated as
Batch System.
follows.

In this study, MATLAB/Simulink is used to create a dynamic model P = Pmax - RI 2 (6)

of an EEMB Li-ion 18650 battery. Fig. 2 shows the overall dynamic
model of a Simulink LIB. A fixed resistance constant was applied where Pmax corresponds to a maximum value of P in the battery
(R = 1.1 Ω), with a simulation time of 2000 s. model.
A. Current, Voltage, and Power Profile Comparison of Constant
It is vital to note that the battery’s power ranges between a peak of
Current and Constant Power Discharge
16.90 W and a constant output of 14.50 W during both the charging
Fig. 6 shows the discharge curves of the EEMB Li-ion 18650 battery
and discharging processes. This indicates an approximate capacity
pack model. The discharge curves of current, voltage, and power
fading of 14%. The conclusion is also influenced by the mechanical
follow the same trend. Two regimes are observed in the three
properties.
graphs: a drop and a relaxation regime. The internal resistance of the
battery caused the voltage drop.

After the application of a constant resistance (R = 1.1Ω), the bat-

tery model under investigation successfully supplied current to the
battery pack connector. Fig. 6a-c depict the current, voltage, and
power characteristics of the battery model during a 2000-s opera-
tion period. In Fig. 6a, it is evident that the current exhibits a gradual
decline over time as the battery undergoes discharge, eventually
reaching a stable state at t = 1800 s. This decrease in current is due
to energy dissipation in the resistor, which generates heat and, as a
result, reduces total current flow, stabilizing the system. In Fig. 6b, it
can be observed that the voltage of a fully charged battery follows
a similar trend as the discharge current, decreasing after surpassing
the rated voltage until t = 1800 s. Given that the discharge and volt-
age curves exhibit a similar pattern, it is imperative to examine the
power generated by the poles of the battery model while consider-
ing the impact of resistive effects.

The power curve exhibits a gradual decrease until it reaches

t = 1600 s, at which point it becomes constant, as depicted in Fig. 6c.
The power output at maximum charge was initially measured at
16.90 W but subsequently decreased to 14.5 W. The resistive effect
of the system is responsible for controlling the spill current. Ohmic
Fig. 6. (a) Current, (b) voltage, and (c) power profile comparison of
power loss exhibits an inverse relationship with resistance, as stated
constant current and constant power discharge.
by relationship in (5).
 TEPES 2024 [epub ahead of print]
Theodore and Şahin. Modeling and Simulation of a Series and Parallel Battery Pack Model

A. Validation of Battery Pack Model

TABLE III.
In the present study, a Li-ion 18650 battery manufactured by EEMB
DIFFERENT SIMULATION PARAMETERS OF LI-ION BATTERY MODEL
AND THEIR VALUES FOR DIFFERENT VALUES OF RESISTANCE was utilized as the battery model. This particular battery has a nomi-
nal voltage of 3.7 V and a capacity of 11.30 Ah. In this study, the
Resistance State of Output Output Power parameters are extracted from a previous study [39], in which the
(Ohms) Charge (%) Current (A) Voltage (V) Delivered (W) resulting experiment demonstrated that the percentage error from
0.5 99.81 8.503 4.252 36.15 the nominal voltage was below 2%. The battery pack model pro-
posed in this study is subsequently validated through a series–paral-
1 99.99 4.287 4.287 18.37 lel arrangement of battery cells. The battery pack model suggested
1.5 99.94 2.866 4.298 12.32
in this study is then validated using a series–parallel configuration of
battery cells. The battery pack model is used to compare simulation
2 99.95 2.152 4.305 9.264 results to the previously reported simulation findings for the single-
battery model.
2.5 99.96 1.723 4.308 7.424

3 99.97 1.437 4.311 6.194 B. State of Charge (%) Versus Time

The SOC of a cell indicates the amount of its current capacity com-
3.5 99.97 1.232 4.312 5.313
pared to its nominal capacity. Fig. 7 illustrates the SOC graph for
both the single-battery and pack-battery models. It can be observed
The determination of the energy expended by the battery through- that the SOC of the single battery shows a straight line indicating its
out the charge/discharge cycle necessitates the integration of the quick discharge, as seen in Fig. 7a, and the SOC graph for the bat-
power function with time. To ascertain the capacity of a battery pack tery pack model gradually discharges from the fully charged state
to provide the desired energy at a particular voltage, Equation (7) to the steady state at T = 10 s, as seen in Fig. 7b. In this simulation,
is employed: the thermal and mechanical impacts on the storage system are not
considered. The SOC of the single-battery model was 89.33% and
W 93.99% for the pack-battery model after applying a load of 0.005 Ω.
C= (7)
3600.U The simulation of the proposed configurations demonstrates that

where C is the capacity of the battery pack (Ah), W is the the total
energy required for the charge/discharge cycles (J), and U is the
nominal voltage of the battery pack (V).

When computing the energy of a simulated charge–discharge cycle,

the sum of the power functions over time gives the total electrical
energy stored in the battery (WTOTAL = 53.9 kJ). LIBs have a capacity
ranging from 25% to 95%, with only 70% of their theoretical capacity
being safe to use. To obtain the right amount, divide the calculated
total energy that the battery should give by (7) [39]. This suggests
that the theoretical required battery capacity is around 5.78 Ah.

As shown in Table III, power output decreases as load resis-

tance increases. If the load resistance drops too much, the
voltage across the internal resistance of the power supply will
drop significantly, greatly impacting performance. However,
the initial charging state was reached when the load resistance
was 1Ω.

VI. PROPOSED BATTERY PACK CONFIGURATION

Battery pack references find application in numerous scenarios that
involve series and parallel cell configurations. Parallel connections
have the effect of increasing battery capacity, while series connec-
tions have the effect of increasing battery voltage. Fig. 2 illustrates
the battery model that is recommended, which is based on a con-
figuration of 26 cells. Thirteen series Li-ion 18650 batteries are
arranged in a configuration where 13 parallel Li-ion 18650 batteries Fig. 7. (a) SOC vs. time curve for the single-battery model and
are combined to create a battery pack model. (b) proposed battery pack model. SOC, state of charge.
 TEPES 2024 [epub ahead of print]
Theodore and Şahin. Modeling and Simulation of a Series and Parallel Battery Pack Model

the discharge rate is lower compared to that of the single-battery

model. This statement signifies the augmentation of its capacity. The
battery cell assembly exhibits a discharge rate of 85.54 A, which is
nearly twice the standard value of 42.77 A for the battery model,
owing to its parallel configuration. In the context of battery systems,
when configured in series, there is no substantial difference in SOC
between the battery model and the series–parallel setups. The bat-
tery pack model simulates SOC with great accuracy, as the observed
inaccuracy is small.

This result was similar to that found by Bhagat et al [20] in their pro-
posed circuit for discharging LIB. They used three Li-ion cells (each
12 V and 6 Ah) connected in series and adapted the nominal volt-
age, rated capacity, and SOC to their needs. A load resistance was
added, and resistance value was set. A metal​-oxid​e-sem​icond​uctor​
field-effect transistor (MOSFET) (to control the circuit) and the cur-
rent measurement were inserted. A scope and a display were added
that showed current readings. The MOSFET was connected to the Fig. 8. Voc–SOC curves of the single Li-ion battery and the battery
pack model obtained at RT. SOC, state of charge.
first battery, and the resistance was connected to the third battery.
Thus, the MOSFET was used as an adjustable resistance. A bus selec-
tor was finally connected to the first battery; the selection in the bus battery pack is roughly double that of a single battery, and the cur-
chooses SOC, current, and voltage. Another scope was inserted to rent and voltage levels remain relatively consistent during the charge
show the waveforms, and the SOC signal was connected as input and discharge operation. The simulation parameters for the OCV and
to the relational operator. The SOC signal was used as an input for SOC values obtained for the single Li-ion and Li-ion pack models are
the relational operator. As a result, when SOC equals zero, the rela- in Table IV.
tion operator outputs a low signal. Voltage measurement connected
its two ends to the positive and negative ends of the battery, fur- Fig. 8 shows the variation of the mean OVC with the SOC for the
ther adding a scope and display to it, including a display of the SOC single Li-ion battery and the battery pack model obtained at 25°C.
value. Although the voltage and current initially fell but eventually It is noteworthy that the battery model will experience depletion
remained constant across the duration, the battery’s SOC reduced before reaching its rated voltage. The SOC curve of the lithium-ion
over time. pack model was consistently found to be associated with the SOC
curve of the battery model. A single battery exhibits a highly linear
The results also showed how battery 3 (12V and 6Ah) and chan-
nel lining evolved. As we can see in this model, the battery’s SOC
decreases with time, and the voltage and current initially decrease TABLE IV.
but then follow a steady pattern throughout the period. The channel THE SIMULATION PARAMETERS OF THE V OC AND SOC OBTAINED
lining remains unaltered during the run. FOR THE SINGLE LI-ION AND LI-ION PACK MODEL

Lithium-ion battery Lithium-Ion Battery Pack

C. Comparison of State of Charge in a Battery Model and Battery
Pack Model Open Circuit State of Open Circuit State of
The OCV curve was defined in our study as the mean value of the Voltage Charge(%) Voltage Charge(%)
charge and discharge equilibrium potentials. OCV is the terminal 2.71538 100 5.93802 100
voltage of the battery in an equilibrium state [40]. The effect of hys-
teresis was overlooked. Furthermore, as stated by reference [41], the 2.54546 97.91188 2.69218 97.91188
OCV–SOC curve is found to be unique for the same type of cell under 2.43682 96.90312 1.42368 96.90312
the same testing conditions. Fig. 8 illustrates the mean OCV at 25°C.
2.35896 96.319 0.92402 96.319
A relatively level gradient ranging from 94% to 95% SOC is empha-
sized in both graphs. The pack model exhibits a significant increase in 2.24854 95.90193 0.72764 95.90193
comparison to that of a single battery. This observation emphasizes
2.20602 95.55058 0.65018 95.55058
the variations in OCV–SOC curves within the range of 93%–100%
SOC at ambient temperatures. It is logical to conclude that the dis- 2.20602 95.22529 0.61915 95.22529
charge capacity of a single-battery model decreases significantly, 2.1689 94.91059 0.60618 94.91059
dropping as much as 95% before stabilizing. Firstly, the battery pack
underwent a complete charging process with a continuous current of 2.13601 94.60049 0.60021 94.600049
1 C. The 1 C rate signifies that the battery would take approximately 2.10658 94.29265 0.59694 94.29265
1 h to discharge completely. The charging process continued until
2.08008 93.98618 0.5947 93.98618
the voltage reached the cutoff value of 5.93 V. The initial value of the
 TEPES 2024 [epub ahead of print]
Theodore and Şahin. Modeling and Simulation of a Series and Parallel Battery Pack Model

curve, which enables an accurate evaluation of its SOC. The SOC can The depiction of the second stage of battery voltage can also be
be pretty accurately approximated using the observed voltage. We observed in Fig. 9b. A phase drop is observed when there is a change
can see that SOC is a superior statistic to voltage for determining in the current level. Another factor to consider is the duration of
how full a specific battery is. This must be considered while design- the forced regime period during which the battery current does not
ing a system, identifying batteries, and using them. reach zero (t = 1600 s). This result was also similar to that of the
discharging circuit of the Li-ion model proposed by Bagh and his
The LIB pack, on the other hand, exhibits a relatively flat discharge team [20] when they studied the variation of current, voltage, and
curve, indicating that the voltage at the battery terminals experi- SOC against time. They discovered that the individual variation of
ences only minimal fluctuations over a wide range of operating the SOC, voltage, and current of a 12V and 6 Ah battery reduced
conditions. The battery input values were utilized to validate the linearly with time. The voltage originally increased from 14 V to 13.8
proposed model in MATLAB, as depicted in Fig. 8. V at t = 0, then began declining parabolically until time t = 270 s,
and remained steady until the conclusion. Likely, in this scenario, the
D. Current, Voltage, and Power Profile Comparison of Constant current increased from 0 to 85.54 A at t = 0, then began to decrease
Current and Constant Power Discharge until t = 1600 s, and remained constant over the entire duration.
The discharge test depicted in Fig. 9 demonstrates that the output
parameters of the proposed battery cell model are higher than those The battery model is first charged to a specific current of I = 85.54 A
achieved with the single-battery model. and a voltage of 94.10 V. It is then discharged in a forced mode until
the relaxation time is achieved at t = 1600 s. The system consistently
Fig. 9a illustrates the observed behavior of the battery model upon provided a power output of 8049 watts, as depicted in Fig. 9c. By
battery drain. The battery cell design was charged to 100% SOC using comparison, the previous battery model had a capacity of 16.85 W
the constant current, constant voltage method. Subsequently, it was and was discharged until reaching 0 V. However, the voltage during
discharged to 0% SOC with a constant current of 1 C. This period is discharge decreases due to the internal resistance of the cell. During
characterized by the cessation of current flow after discharge, and the process of power discharge, the voltage of the cell remained
the battery voltage gradually reaches a steady state. The battery cell consistently at 36.02 V, leading to a corresponding current value
model demonstrates a discharge in forced mode until t = 1600 s, fol- of 30.44 A. The calculation of the total energy consumption follow-
lowed by a transition to relaxed mode. During the idle phase, the ing complete discharge yields a value of WTOTAL = 16.12 MJ, which is
processes of polarization and material exchange within the battery three orders of magnitude greater than the 53.9 KJ estimated in the
exhibit sluggishness, resulting in voltage fluctuations. In comparison battery pack. The pack model exhibited a capacity of 17.28 mAh.
to the battery of the conventional model, it was observed that the
maximum current value (I) reached 42.77 A and the maximum volt- E. Instantaneous Discharge Power Versus State of Charge
age measured was 94 V. These values were found to be higher than The discharge performance was assessed in the proposed model,
those of a single battery, indicating an increased capacity. with the SOC being considered a variable. This evaluation is depicted
in Fig. 10. In the present study, the resistance of the battery (Rbatt)
and the battery’s SOC are considered variables dependent on the
temperature. The assumption is made that the battery temperature
remains constant at 20°C and any potential temperature effects are
disregarded.

The instantaneous current capacity of the battery pack is depicted

in Fig. 10. The text provides a characterization of battery models uti-
lized in electrical systems, specifically in the context of hybrid electric
vehicles (HEVs). Standard experimental procedures were accessible
within the advisor software utilized to generate the SOC chart data
for the batteries employed. The proposed battery design demon-
strates applicability despite the individual battery cell having a mass
of 38 g and a nominal voltage of 3.7 V. This is due to its ability to
be discharged up to 88 times, with a charge–discharge efficiency of
80%–90% for LIBs.

The performance of the battery model has exhibited significant

improvement compared to the conventional model. A twofold
increase in current output and a thirteenfold increase in voltage
output were observed as a result of employing the 13-parallel and
13-series battery configurations, respectively. The battery pack
model demonstrated exceptional electrical performance at ambient
Fig. 9. (a) Current, (b) voltage, and (c) power discharge profile
comparison of constant current and constant power discharge. temperatures, surpassing that of a single battery with better SOC
stability after a long discharge process. The observed characteristics
 TEPES 2024 [epub ahead of print]
Theodore and Şahin. Modeling and Simulation of a Series and Parallel Battery Pack Model

or Interpretation – T.A.M.; Literature search – T.A.M., Ş.M.E.; Writing – T.A.M.;

Critical Review – Ş.M.E.

Declaration of Interests: The authors have no conflict of interest to declare.

Funding: This research received no funding.

REFERENCES
1. A. M. Theodore, S. Konwar, P. K. Singh, R. M. Mehra, Y. Kumar, and
M. Gupta, “PEO + NaSCN and ionic liquid-based polymer electrolyte
for supercapacitor,” Mater. Today Proc., vol. 34, no. 3, pp. 802–812,
2021.
2. N. Badi et al., “The impact of polymer electrolyte properties on lithium-
ion batteries,” Polymers, vol. 14, no. 15, pp. 3101, 2022. [CrossRef]
3. ”N. Badi, A. M. Theodore, A. Roy, S. A. Alghamdi, A. O. M. Alzahrani,
and A. Ignatiev, “Preparation and characterization of 3D porous silicon
anode material for lithium-ion battery application,” Int. J. Electrochem.
Sci., vol.17, no. 6, pp. 22064, 2022. [CrossRef]
Fig. 10. Instantaneous discharge power versus SOC. SOC, state of 4. A. M. Theodore, “Progress into lithium-ion battery research,” J. Chem.
charge. Res., vol.47, no.3, pp. 1–9, 2023. [CrossRef]
5. A. M. Theodore, “Promising cathode materials for rechargeable lithium-
ion batteries: A review,” JSE, vol. 14, no. 1, pp. 51–58, 2023.
of current, voltage, and power discharge demonstrate the antici- 6. S. Wang, C. Lu, C. Liu, Y. Zhou, J. Bi, and X. Zhao, “Understanding the
pated enhancement in the performance of the proposed battery cell energy consumption of battery electric buses in urban public transport
systems,” Sustainability, vol. 12, no. 23, pp. 10007, 2020. [CrossRef]
model, indicating its potential applicability to various battery types.
7. X. Y. Mao, “Research on Li-ion Battery State of Charge and Active Equali-
zation Technology,” Master’s Thesis. Nanjing, China: Nanjing Univ. Posts
In this paper, the temperature effect is neglected. Taking the tem- Telecommun., 2020.
perature effect into account necessitates the modeling effort men- 8. M. Chen, and G. A. Rincon-Mora, “Accurate electrical battery model
tioned in [42]. Using a 3D lookup table, the temperature effect can capable of predicting runtime and I-V performance,” IEEE Trans. Energy
be included with a few adjustments. Convers., vol. 21, no. 2, pp. 504–511, 2006. [CrossRef]
9. C. Liao, H. Li, and L. Wang, “A dynamic equivalent circuit model of
VII. CONCLUSION LiFePO4 cathode material for lithium-ion batteries on hybrid electric
In this paper, it is aimed to provide guidance and simulate a model vehicles,” IEEE Vehicle Power and Propulsion Conference, 2009, pp.
of the EEMB Li-ion 18650 battery pack. The battery model proposed 1662–1665. [CrossRef]
in this study is configured in a series–parallel arrangement, allowing 10. W. Y. Low, J. A. Aziz, N. R. N. Idris, and R. Saidur, “Electrical model to
predict current-voltage behaviors of lithium ferro phosphate batteries
for the simulation of various discharge currents and the assessment
using a transient response correction method,” J. Power Sources, vol.
of their impact on battery performance. Additionally, the model can
221, pp. 201–209, 2013. [CrossRef]
estimate the battery’s SOC and output power. The SOC of a series– 11. M. C. Knauff, C. J. Dafis, D. Niebur, H. G. Kwatny, C. O. Nwankpa, and J.
parallel Li-ion battery is modeled and simulated using the MATLAB/ Metzer, “Simulink model for hybrid power system test-bed,” IEEE Elec-
Simulink tool. State of charge is investigated to determine the dis- tric Ship Technologies Symposium, Arlington, VA, USA, 2007, pp.
charging behavior of a Li-ion battery pack. When a load is supplied to 421–427. [CrossRef]
the battery, the performance of the suggested model, which includes 12. ”K. Yusuf, A. Ö. İlker, and S. İsmail, “Estimation of li-ion battery state of
SOC, current, voltage, and power, is evaluated. Modeling and simu- charge using adaptive neural fuzzy inference system (ANFIS),” IJEAT, Vol.
lating the circuit with set charge and discharge percentages aids 7, no. 3, 2020, pp. 88–94.
in improving battery life and battery degradation, as shown in the 13. A. Afzal, R. K. Abdul Razak, A. D. Mohammed Samee, R. Kumar, Ü.
Ağbulut, and S. G. Park, “A critical review on renewable battery thermal
battery circuit, has an SOC (%) that remains relatively constant with
management system using heat pipes,” J. Therm. Anal. Calorim., vol. 2,
varying loads, indicating the model’s stability at ambient tempera-
pp. 1–40, 2023. [CrossRef]
ture (Table III). The proposed model outperforms the single-battery 14. K. Jaouad, B. Najib, and S. Abdallah, “Review on lithium-Ion battery
model significantly. This study aims to contribute to the analysis of modeling for different applications,” IJEAP, vol. 1, no. 1, pp. 38–47, 2021.
HEV battery capacity by predicting the discharge characteristics at 15. M. Urbain, “Modélisation Électrique et Énergétique des Accumulateurs
various battery capacities. The simulation results demonstrate a Li-Ion. Estimation en Ligne de la SOC et de la SOH,” Thèse de doctorat.
strong resemblance between the proposed simulation model and Institut National Polytechnique de Lorraine, 2009.
the recorded test results from the battery data. 16. F. Miomandre, S. Sadki, P. Audebert, and R. Méallet-Renault, “Électro-
chimie - 2e éd.:Des concepts aux applications,” Dunod, 2011. Available:
https​://bo​oks.g​oogle​.com.​tr/bo​oks?i​d=zis​0_5G_​x4sC
Peer-review: Externally peer-reviewed. 17. A. Eddahech, Modélisation du vieillissement et détermination de l’état
de santé de batteries lithium-ion pour application véhicule électrique et
Author Contributions: Concept – T.A.M.; Design – T.A.M.; Supervision – hybride, Université Sciences et Technologies - Bordeaux I, 2013. Fran-
Ş.M.E.; Ressources – T.A.M., Ş.M.E.; Materials – T.A.M., Ş.M.E.; Analysis and/ çais. (NNT : 2013BOR14992). (tel-00957678)
 TEPES 2024 [epub ahead of print]
Theodore and Şahin. Modeling and Simulation of a Series and Parallel Battery Pack Model

18. S. Mahmud et al., “Recent advances in lithium-ion battery materials for 30. T. Huria, M. Ceraolo, J. Gazzarri, and R. Jackey, “High fidelity electrical
improved electrochemical performance: A review,” Results Eng., vol. 15, model with thermal dependence for characterization and simulation of
pp. 100472, 2022. [CrossRef] high-power lithium battery cells,” IEEE International Electric Vehicle
19. L. Ianniciello, P. H. Biwolé, and P. Achard, “Electric vehicle batteries ther- Conference, Greenville, SC, USA, Vol. 2012, 2012, pp. 1–8. [CrossRef]
mal management systems employing phase change materials,” J. Power 31. A. J. Salkind, C. Fennie, P. Singh, T. Atwater, and D. E. Reisner, “Determi-
Sources, vol. 378, pp. 383–403, 2018. [CrossRef] nation of state-of-charge and state-of-health of batteries by fuzzy logic
20. S. Bhagat et al., “Simulation of Li-ion Battery using MATLAB Simulink methodology,” J. Power Sources, vol.80, no.1–2, pp. 293–300, July 1999.
for Charging and Discharging,”E3S Web of Conferences 353, 03001, [CrossRef]
EVF’2021, 2022. 32. L. David, and B. R. Thomas, Handbook of Batteries, 3rd ed., 1995.
21. Q. Wang, B. Jiang, B. Li, and Y. Yan, “A critical review of thermal manage- 33. G. Marcia, “A battery’s guide to immortality,” 2019. Accessed: 08 August
ment models and solutions of lithium-ion batteries for the development 2019. Available: https​://ww​w.mtu​.edu/​magaz​ine/r​esear​ch/20​19/st​
of pure electric vehicles,” Renew. Sustain. Energy Rev., vol. 64, pp. ories​/ battery-immortality/
106–128, 2016. [CrossRef] 34. H. Xuebing, O. Minggao, and L. Languang, “A comparative study of com-
22. Y. Miao, P. Hynan, A. von Jouanne, and A. Yokochi, “ Current li-ion bat- mercial lithium-ion battery cycle life in the electrical vehicle: Aging
tery technologies in electric vehicles and opportunities for advance- mechanism identification,” J. Power Sources, vol. 251, pp. 38–54, 2014.
ments,” Energies, vol. 12, no. 6, pp. 1074, 2019. [CrossRef] [CrossRef].
23. M. A. Hannan, M. S. H. Lipu, A. Hussain, and A. Mohamed, “A review of 35. Available: https​://ww​w.mat​hwork​s.com​/help​/phys​mod/s​ps/po​wersy​s/
lithium-ion battery state of charge estimation and management system ref​/batt​ery.h​tml.
in electric vehicle applications: Challenges and recommendations,” 36. O. Tremblay, and L. A. Dessaint, “Experimental validation of a battery
Renew. Sustain. Energy Rev., vol. 78, pp. 834–854, 2017. [CrossRef] dynamic model for EV applications,” World Electr. Veh. J., vol. 3, no.2,
24. S. B. Lee, R. S. Thiagarajan, V. R. Subramanian, and S. Onori, “Advanced pp. 289–298, 2009. [CrossRef]
battery management systems: Modeling and numerical simulation for 37. K. W. E. Cheng, B. P. Divakar, H. Wu, K. Ding, and H. F. Ho, “Battery-
control,” Proceedings of the 2022 American Control Conference Management System (BMS) and SOC development for electrical vehi-
(ACCC), Atlanta, GA, USA, 8–10 June, 2022, pp. 4403–4414, 2022. cles,” IEEE Trans. Veh. Technol., vol. 60, no. 1, pp. 76–88, Jan. 2011.
[CrossRef] [CrossRef]
25. U. Krewer, F. Röder, E. Harinath, R. D. Braatz, B. Bedürftig, and R. Find- 38. D. Shen, L. Wu, G. Kang, Y. Guan, and Z. Peng, “A novel online method
eisen, “Review—Dynamic models of li-ion batteries for diagnosis and for predicting the remaining useful life of lithium-ion batteries consider-
operation: A review and perspective,” J. Electrochem. Soc., vol. 165, no. ing random variable discharge current,” Energy, vol. 218, pp. 119490,
16, pp. A3656–A3673, 2018. [CrossRef] 2021. [CrossRef]
26. F. Brosa et al., “A continuum of physics-based lithium-ion battery mod- 39. I. Husain, “Electric and hybrid vehicles: Design fundamentals,” Front
els reviewed,” Prog. Energy, vol. 4, no. 04, 2003. Cover. CRC Press, - Technology & Engineering, pp. 288, 2003.
27. M. Martí-Florences, A. Cecilia, and R. Costa-Castelló, “Modelling and 40. L. W. Yao, J. A. Aziz, P. Y. Kong, and N. R. N. Idris, “Modeling of lithium-
estimation in lithium-ion batteries: A literature review, Martí-Florences, ion battery using MATLAB / Simulink,” IECON, vol. 2013, pp. 1729–1734,
A. Cecilia, R, Energies, Vol.16, no.19, 2023, pp. 6846. 2013. [CrossRef]
28. K. Movassagh, A. Raihan, B. Balasingam, and K. Pattipati, “A critical look 41. H. Dai, X. Wei, Z. Sun, J. Wang, and W. Gu, “Online cell SOC estima-
at coulomb counting approach for state of charge estimation in batter- tion of Li-ion battery packs using a dual time-scale Kalman filtering
ies,” Energies, vol. 14, no. 14, pp. 4074, 2021. [CrossRef] for EV applications,” Appl. Energy, vol. 95, pp. 227–237, 2012.
29. P. Singh, and D. Reisner, “Fuzzy logic-based state-of-health determina- [CrossRef]
tion of lead-acid batteries,” 24th Annual International Telecommunica- 42. L. Long, P. Bauer, and E. Kelder, “A practical circuit-based model for Li-ion
tions Energy Conference, Montreal, QC, Canada, 2002, pp. 583–590. battery cells in electric vehicle applications,” IEEE Telecommunications
[CrossRef] Energy Conference (INTELEC), 2011, pp. 1–9. [CrossRef]